Options for bilingual web content and navigation

We've recently loaded up a number of documents from the www.natlib.govt.nz redevelopment project to share... redesign strategy reports, RFPs, CMS requirements, project task lists, usability study plans, Zope/Plone performance test results etc. Among these docs is one particular gem that I want to draw out... the options we looked at for presenting web content in both English and te reo Māori.

I'll jump straight into the options we looked at. If you've used or seen other approaches, or have got your own views on these options, please add a comment.

Option one - Duplicate content

The approach:

• One site
• Pages within the site have content in both languages
• No ability to switch between languages; both are always present.

The pros:

• Displays both languages together
• An extreme interwoven option
• Raises awareness of te reo
• Challenges us to commit to producing te reo content.

The cons:

• Significant technical challenges (search, RSS, screen readers)
• Content doesn't exist and might not exist; a big challenge
• Users would mistakenly assume that it is an exact translation
• Serious user issues (e.g., page length)
• What if there is no other language?
• Design issues if more text in one language
• Some issues addressed in content doesn't equate; stories are told differently
• Makes story-telling harder
• How would we display three languages?
• Increased maintenance; vastly increased $ for content production

Option two - Mirror IA with language switch

The approach:

• Site provided in two languages
• IA exactly the same
• Switch allows user to go to current page in alternate language's site.

The pros:

• Good experience for fluent speakers, if all content is translated
• Forces us to think about providing te reo content.

The cons:

• Requires a complete translation
• Workload and workflow issues
• Would require development of existing content
• Not an interwoven solution
• There is a lack of political driver in NZ (compared to the Canadian model, for example)
• The user expects that the site will be fully translated, and that they will always get content in te reo Māori if they click on a Māori link.

Option three - Content area language switch

The approach:

• Page mainly in one language, with an area of content in the other
• User can 'switch' the content area that is in the other language, so that the page is fully in one language
• Rest of the page unaffected
• Rest of site unaffected.

The pros:

• Good for descriptions/instructional content
• Somewhat interwoven
• Allows user to contextualise (even when the other content isn't translated)
• Allows key-word and full-text search on both languages
• Clear about what is a translation and what is not.

The cons:

• Not fully bilingual
• Blue users might miss content in the green language, if that content is not also provided in blue on the page originally displayed.

Option four - Special content paths

The approach:

• One site; one IA
• Most pages in one language
• Some content/pages in the other language
• Special content
• Unique paths.

The pros:

• Allows stories to be told in unique way for each language
• Good if content really can be told in one way.

The cons:

• Inaccessible to users who do not speak both languages
• Not really interwoven, unless you are fully bilingual

Option five - Mirror content with language switch

The approach:

• Page mainly in one language, with an area of content in the other
• User can 'switch' the content, so that the page is mainly in the other language, with an area of content in the original
• Rest of site unaffected; switch only applies to the page.

The pros:

• Interwoven
• Supports lots of te reo Māori content
• Lays down the challenge to actively develop content
• Could work with other secondary languages
• Site is immediately fully bilingual.

The cons:

• For 'blue' readers, the 'green' information might be frustrating; need to toggle to know if it is relevant
• All content must be available in two languages; not available at present
• Seems to use a direct translation; doesn't allow for different ways of telling stories in different languages
• Anyone operating in the green language would be initially lobbed into the blue language; switch would be frustrating
• High maintenance
• Not clear from the home page that the green language is equally accessible.

Option six - Separate sites

The approach:

• Two sites; one in English and one in te reo Māori
• Home page gives access to both sites
• Sites are different; not the same IA or content
• Back to the homepage to change language.

The pros:

• Ability to develop different ways of searching and presenting information
• Could be Māori-centric
• Would challenge us to develop content in te reo Māori.

The cons:

• How would we deal with more than 2 languages
• Not interwoven; one experience or the other, not both
• Content differs; need to be clear about this is a 'green' user gets less information than the 'blue' user
• May not be appropriate to house taonga but use a 'blue' language alone
• We don't think we have enough content in both languages
• High maintenance cost; two sites
• Not supportive of use of te reo Māori.

What we decided?

For our situation, we chose to implement a variation on option five: mirror content with language switch. A good example of this is action is our Māori resources collection page. You'll see that when you switch to the te reo version, only the content area is translated, the navigation items surrounding the core content stay in English. The variation is that only the summary at the top is translated on some pages, and you can't switch to a full translation e.g. our introduction to digitisation page.

The cost factor of translating every single page during the project was significant for us. This solution allows us to translate more pages as we get budget. Most pages have the summary box translated at the top, and we have about 30% of static content (i.e not events, job, or news) translated so far. Once we have translated a critical mass of pages into te reo our plan is to translate the IA and navigation and allow full bilingual switching. This would be similar to option two - mirror IA with language switch - but would include the interwoven summary boxes. We'll have to do more user testing at the time though. Our goal is to eventually have a fully bilingual site, and we think this current approach is a good stepping stone for us and our users.

Thanks to Shift!

There were a number of vendors involved in the overall redevelopment project, but it was Shift who facilitated this particular piece of work. Full credit to them, we found this particular piece of analysis very useful.